Increasing Performance with VIEWS

Hi,

I’m working on some blogging software which uses MySQL as a datas tore.

Each archive page, year, month, day can have a tag cloud generated to show the most popular tags for that day.

Currently I’m using a derived table. Then I stumbled across your article “Derived Tables and Views Performance” and tested a query with a VIEW and it was indeed theoretically faster because the number of rows required in the query that used the VIEW was SUBSTANTIALLY lower than that of the derived query.

What I was thinking of implementing was temporay VIEWs each time the tag cloud needs to be generated (every page view eek: ).

So it would go:

CREATE VIEW for IDs

PERFORM query USING VIEW

DROP VIEW

Remembering this would be performed every page view, for every visitor on every blog hosted (we’re talking a lot of VIEWS being created and DROPPED here) - how would it perform? Does this require a lot of IO operations?

Thank you!

Why ask, better benchmark it )

Generally it hould not be that expensive however if --sync-frm is dependent it may need fsync which is relatively expensive.